[A-AS08] Processes of the Moist Atmosphere Across Scales

[AAS08-04] Impact of implicit numerical diffusion on mean states in a super-parameterized model

*Kazuya Yamazaki¹ (1.Information Technology Center, the University of Tokyo)

Keywords:Atmospheric model, Super-parameterization

Super-parameterization is an atmospheric modelling technique in which a global climate model (GCM) is coupled to numerous domains of a high-resolution cloud-resolving model (CRM). The CRM simulates small-scale activities such as cumulus convection and conveys their impact to the GCM, behaving as a rich parameterization scheme for the GCM.
One can couple a CRM domain to either one column or a block of multiple columns of the GCM. Experiments using SP-MIROC, a super-parameterized model consisting of MIROC6 and SCALE-RM, suggested that coupling CRM domains to blocks of GCM columns, instead of individual ones, mitigates SP-MIROC's cold bias in the upper troposphere.
When CRM domains are coupled to individual GCM columns, implicit numerical diffusion in the GCM excessively dissipates heat generated in the CRM domains, behaving similarly to reinforced stratification. This effective over-strafitication gets offsetted by the weakening of the actual tropospheric stratification in long terms, resulting in a cold bias in the upper troposphere. Coupling CRM domains to blocks of columns makes the them less sensitive to the GCM's excessive implicit diffusion, allowing the super-parameterized model to simulate more realistic temperature profiles.